Project Summary Many important cellular processes, such as gene expression regulation, transport, and cell division, are carried out by protein complexes. To compensate for the difficulty of solving protein complex structures, computational approaches for protein docking prediction have been developed over recent decades. Although steady progress has been observed in the field, current methods and developments are largely limited to pairwise protein docking. The proposed project aims to extend the capability of protein docking methods to two important classes of protein complexes, multimeric complexes and disordered interactions, while making further improvements in pairwise protein docking. A substantial fraction of protein complexes involved in critical cellular processes are multimeric complexes or involve interactions with intrinsically disordered regions. As these two types of complex structures are particularly difficult to determine by experimental means, it is an urgent and important task for protein structural bioinformatics to develop efficient and accurate computational methods to build models for these types of complexes. Structure models provide hypotheses for designing biochemical experiments to elucidate interactions in a complex, which can lead to solving full or partial complex structures. The structural information provided by multimeric protein docking is not merely an incremental scale-up of pairwise docking; multimeric docking can also predict interacting and non-interacting subunits and assembly order within a complex. The developed docking methods will be applied to build models of protein complexes from four biological systems in collaboration with biologists. The proposed methods for protein complex structure prediction are also useful for designing drugs for protein-protein interaction targets as well as artificial design of protein complexes and bio- nanomaterials. Collaboration with biologists will further enhance integrated computational and experimental approaches in biology.